Futarchy #4: decision markets derivatives and applied AI
Citing again Robin Hanson, right where he used the slogan “Vote on Values, But Bet on Beliefs”:
In futarchy, democracy would continue to say what we want, but betting markets would now say how to get it. That is, elected representatives would formally define and manage an after-the-fact measurement of national welfare, while market speculators would say which policies they expect to raise national welfare. The basic rule of government would be:
When a betting market clearly estimates that a proposed policy would increase expected national welfare, that proposal becomes law.
Futarchy is intended to be ideologically neutral; it could result in anything from extreme socialism to extreme monarchy, depending on what voters want, and on what speculators think would get it for them.
Futarchy seems promising if we accept the following three assumptions:
- Democracies fail largely by not aggregating available information.
- It is not that hard to tell rich happy nations from poor miserable ones.
- Betting markets are our best known institution for aggregating information.
Interestingly, Hanson was also involved in the creation of the Foresight Exchange, an online play predictions market in which current market prices reflect consensus about the future, and FutureMAP, a (now canceled) DARPA research project into the use of prediction markets for shaping government policy.
By creating a decisions market, we are also allowing for making money out of market imprecisions, and therefore helping correct them. For example, the market attributes a specific pricing to decision A. That’s public information. Imagine you observe the price, and it looks wrong to you, maybe it’s higher or lower than your expectations. You can now bet against that price, and make money through fixing the market price.
Importantly,
prediction markets cannot only give probability estimates for the effectiveness of policies, but also probability estimates for future indicators chosen by future governments. This is important, because taking into account our future moral values might be different. So we have to be aware that in the future people might have other preferences for their moral values, or that new insights and technologies allow for the adoption of other, better indicators in the future.
(Stijn Bruers, Rational democracy and futarchy)
An entire set of primitives could be generated, just like stocks, to support the market in finding the right prices associated with every decision. The use of well-known instruments proven to optimize speculative markets, such as options and futures, would enable decision markets to create more precise pricing, and therefore reinforce the strength of this new approach. The more value is in playing this game, the more sophisticated players we would attract, ultimately benefiting the precision of the results.
Creating sophisticated markets for assessing decisions means leveraging open markets to access our collective intelligence in an organized and transparent way, to find the decision with the best impact on our future. This is decentralization in its purest form.
As we mentioned at the beginning of the series, futarchy could have a very interesting implementation in an optic where we leverage AI. If futarchy was originally designed as a decision-making procedure for aggregating the intelligence of humans with the purpose of governance, AIs could also be built on futarchies. As discussed in Futarchy implements evidential decision theory, many approaches to AI (i.e. Deep Learning) are organisms where all knowledge is concentrated into a single entity, and they don’t need any aggregating procedure (i.e. democracy or decentralized intelligence or the efficient information aggregation and processing of speculative markets).
However, it has also been proposed that intelligence arises from the interaction and sometimes competition of many simple subagents — see, for instance, Minsky’s book The Society of Mind, Dennett’s Consciousness Explained, and the modularity of mind hypothesis.
Prediction markets would be models that combine the opinions of many of these agents, while functioning in a radically different way compared to a human brain.
A theoretical example of the use of prediction markets in AI is MIRI’s logical induction paper. Furthermore, markets are generally similar to evolutionary algorithms.
As we end this first analysis of futarchic governance, we realize how this approach calls into action several concepts dear to the Web3 community, and allows them to work together in synergy:
- Decentralization of decision making
- Democratic governance
- Optimized, possibly smart-contract based metrics measurement and optimization
- Prediction and decision markets
- Speculative markets with their primitives
This, to me, is utterly fascinating. Let’s continue our exploration by looking at the applications of futarchy across the private and public spheres.
